Story thread

Mistral Le Chonk launch

Mistral is previewing Mistral Large 4, a 1-trillion-parameter open-weight model it says beats other Western open models, with weights due October 27.

8 stories from 5 sources · since Oct 6

Clio’s read on this thread

Mistral has launched or previewed Le Chonk, a 1-trillion-parameter model positioned as a strong open-weight alternative to top US and Chinese systems. The release matters for CIOs because it expands options for sovereign, private, and security-sensitive AI deployments, but it also raises hosting, governance, and infrastructure demands. There is still some confusion over when weights will be fully available and how open the model is.

Key points

  • Le Chonk/Mistral Large 4 is a 1T-parameter multimodal model aimed at agentic and enterprise workloads.
  • Mistral says the model was trained on 3,800-4,000 Nvidia Grace Blackwell GPUs in Europe.
  • Weights are slated for October 27, but TechCrunch says the model is not yet fully open-weight.
  • Third-party Artificial Analysis ranks it as the strongest model outside the US and China, near DeepSeek V4.1 Flash(max).
  • Target use cases include coding, cyberdefense, finance, chip design, and sovereign on-prem deployments.

Where coverage agrees

  • Mistral's release centers on Le Chonk/Mistral Large 4, a trillion-parameter model.TechMeme · Wired · TechCrunch · The Register · Ars Technica
  • The model is being pitched as an enterprise option with more control over deployment, governance, and data residency.TechMeme · Wired · TechCrunch · The Register · Ars Technica
  • The launch is framed as competition with proprietary US frontier models and, in some coverage, open Chinese models too.TechMeme · Wired · TechCrunch · The Register · Ars Technica

Where sources differ

  • One TechMeme report says Mistral trained the model on 4,000 Grace Blackwell GPUs, while another says 3,800 GPUs.TechMeme
  • TechCrunch says Le Chonk is not yet fully open-weight and the weights arrive October 27, while Wired, The Register, and Ars Technica describe it as open-weight now.TechCrunch · Wired · The Register · Ars Technica
  1. Mistral says "Le Chonk" can challenge the best AI modelsArs Technica · Oct 7 · Clio 70
  2. Artificial Analysis says Mistral Large 4 is the most intelligent model from outside the US and China, achieving results comparable to DeepSeek V4.1 Flash (max) (Artificial Analysis)TechMeme · Oct 6 · Clio 63
  3. Mistral launches a preview of Mistral Large 4, or Le Chonk, a 1T model it claims tops any open model developed in the US or Europe; weights are due October 27 (Carl Franzen/VentureBeat)TechMeme · Oct 6 · Clio 70
  4. Mistral says ML4 was trained using 3,800 Nvidia Grace Blackwell GPUs in its own data centers in Europe and much of its training data was multilingual (Mistral Blog)TechMeme · Oct 6 · Clio 70
  5. European AI flag bearer Mistral's new open weights model is 'Le Chonk'The Register · Oct 6 · Clio 69
  6. Mistral’s new 1T model aims to leapfrog closed and open rivalsTechCrunch · Oct 6 · Clio 65
  7. Mistral Says Its New AI Model ‘Le Chonk’ Is the Best Open-Weight Offering Outside of ChinaWired · Oct 6 · Clio 62
  8. Mistral releases Mistral Large 4, dubbed "le Chonk", a 1T-parameter open-weight model for general agentic capabilities, trained on 4,000 Grace Blackwell GPUs (Sabrina Ortiz/The Deep View)TechMeme · Oct 6 · Clio 68